There are lots of reasons for higher failure rates in one discipline than another that have nothing to do with how "scientific" they are. Cancer biology is indi

There are lots of reasons for higher failure rates in one discipline than another that have nothing to do with how "scientific" they are. Cancer biology is indisputably a science, and has a higher rate of non-replicable results than non-organic chemistry; that's because it's much more complicated, not because it's some sort of unscientific philosophizing. Psychology, which gets the most grief from hard physical STEM types whenever this issue comes up, is hard to reproduce because it's hard in general. It's the study of the most complex object we know about, and we can only access it through indirect and creative means. Small differences in study design can have large impacts on outcomes because the brain is sensitive, and complicated, and often counter-intuitive. It's not that there's more "bad science" being done in some disciplines, it's that those disciplines require much more scrutiny towards their methods and inferences, because they're harder to handle. But they're still science, and it would be a horrible disservice to abandon the scientific method for them just because it's hard.